Let's assume you really must build this power supply.  I seem to have been way
off, in that these cars really take that large amount of energy.  I agree with
the others that one should add some form of battery at the output to enable a
high peak current draw.  The call then is simply to make a supply that supplies
most of the power draw, or a fraction, which would have to charge back up with
some delay.  Let us assume you want to supply 13VDC at 40 amps.  That ought to
handle the bulk of the use, barring the high acceleration peaks.  That is a 520W
supply, or thereabouts.  I have gleaned much knowledge from two sources.  One is
the Unitrode (now TI) databook and applications book.  They make discrete PWM
controllers that are very popular in PC switching supplies.  Next, Power
Integrations make a line of lovely off line switchers.  They are nice too.  They
tend to partner with Premier Magnetics (loose use of the term) to make off the
shelf magnetics parts available.  Solves things nicely sometimes.

BUT you need >520W.  No off the shelf magnetics will do that.  Further assume
you want to make this thing work off line, to avoid the use of a 5000 KW linear
supply on the front (changed my mind on that too).  Now we know that the
switcher will convert line voltage down to 13V or so.  The biggest hurdle to
advanced switcher design is the transformer.  Power Integrations actually offers
an Excel spreadsheet for entering parameters and getting design calculations
done for you.  Their databook says so, I assume it is on their web page.  This
could give you the magnetics design parameters.  Then you can find a Power TOPS
big enough, or go to a Unitrode part with an external FET.  There will be a
sizeable 400VDC electrolytic in the rectified line side, and the snubber on the
input side of the transformer will be an apreciable challenge. (leakage on a big
transformer will be relatively high, and so switching waste will be high).  The
output will need some sort of capacitor, but it probably is not critical since
you will have the battery.  The output diode will have to handle some seriious
power.  Remember, 50A ave could be 100A peak or more.  A TO220 part from GI may
be necessary.

The plus is that you can start with the standard feedback circuit for the
switcher in question, and add some circuitry to allow fast charge voltage and
current limits to be custom set to the battery in use.  That way, whe the supply
is not being belted with huge loads, it is properly charging the battery, not
boiling it off.

I would run this thing at 80 to 100KHz.  Anything higher would turn into a NASA
project.

And make certain that you achieve very effective line isolation in the layout
and opto feedback circuits.  It will not do your reputation any good if you fry
a whole troop of Cub Scouts.

Are we going in the right direction yet?  Do you still want to do this?
Chris Eddy